DiagNet replaces the object-detection neck with a GCN trained against a diagonal-constraint target on the adjacency matrix, claiming mAP improvements over YOLO baselines, though the experiments are not reproducible and the baseline comparisons are suspicious.
K., Winn, J., and Zisserman, A
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
DiagNet: Detecting Objects using Diagonal Constraints on Adjacency Matrix of Graph Neural Network
DiagNet replaces the object-detection neck with a GCN trained against a diagonal-constraint target on the adjacency matrix, claiming mAP improvements over YOLO baselines, though the experiments are not reproducible and the baseline comparisons are suspicious.